Prognostic performance of radiomic data to predict distant metastases and overall survival in patients with early stage (I/II) non-small cell lung cancer (NSCLC)

肺癌是世界上最主要的癌症死亡原因，非小细胞肺癌(NSCLC)是最常见的亚型。尽管在诊断和治疗方面有所改进，但非小细胞肺癌的预后仍然很差。只有早期疾病的患者才能在手术切除的基础上进行治疗。然而，即使在完全和成功地切除了相同的肿瘤阶段，也观察到了广泛的生存期。在这种背景下，越来越明显的是，每个肿瘤都与不同的表型特征有关，这些表型特征导致了不同的治疗反应。这导致了个性化医疗的概念，目的是根据患者的个人需求量身定做治疗策略。因此，个体肿瘤表型可以作为评估个体对不同治疗方法的反应的决策工具，目的是改善总体生存和生活质量。肺癌的常规诊断是使用医学成像，最常见的是计算机断层扫描(CT)。诊断是基于放射科医生根据不同的成像特征进行的主观图像解释，具有由于观察者之间的可变性而导致的误诊风险。因此，肿瘤组织的活检对于确定最终诊断是必要的。然而，活组织检查只允许获取小样本的组织，这可能无法全面捕捉肿瘤内的分子变化，因此存在低估肿瘤侵袭力的风险。因此，基于数学算法的非侵入性提取客观图像特征来描述肿瘤表型的定量成像分析变得越来越重要。这种分析成像数据的新方法被称为放射组学。越来越多的证据表明，放射组学通过提供有关肿瘤分期、转移潜能、治疗反应和总生存期的预后信息，有可能成为个性化医学的一种低成本、非侵入性的生物标志物。早期非小细胞肺癌患者特别适合于评估放射组学对非侵袭性肿瘤侧写的预后影响，因为这些患者通常可以获得CT图像和组织病理学分析。此外，如上所述，尽管手术切除了相同阶段的疾病，但众所周知但到目前为止还不能预测，这些患者中的某些亚群将经历复发的疾病，这表明需要精确的药物治疗。因此，本研究的目的是调查早期非小细胞肺癌患者的放射学数据的预后表现，目的是确定预测性的影像特征，以估计转移潜能和总生存率。这不仅对个性化医学非常重要，而且从跨学科和社会经济学的角度来看也是如此。

Lung cancer is the leading cause of cancer deaths in the world with non-small cell lung cancer (NSCLC) being the most commonly observed subtype. Despite improvements in diagnosis and treatment, prognosis of NSCLC remains poor. Only patients with early-stage disease are treated with curative intent based on surgical resection. However, even after complete and successful resection of the same tumor stage, a wide spectrum of survival periods is observed. In this context, it has become increasingly evident that each tumor is associated with distinct phenotypic characteristics, which are responsible for different treatment responses. This has led to concept of personalized medicine with the aim to tailor treatment strategies towards the individual needs of the patient. Thus, the individual tumor phenotype may serve as a decision-making tool for estimating the individual response to different treatment approaches with the aim to improve overall survival and quality of life. Lung cancer is routinely diagnosed using medical imaging, most commonly computed tomography (CT). Diagnosis is based on the subjective image interpretation of the radiologist according to distinct imaging features with the risk for false diagnosis due to inter-observer variability. Therefore, biopsy of tumor tissue is necessary to establish a final diagnosis. However, biopsy only allows for acquiring small samples of tissue, which may fail to comprehensively capture the molecular variations within tumors, thus posing the risk for underestimating tumor aggressiveness. Therefore, quantitative imaging analysis has gained increasing importance with the aim to non-invasively extract objective image features to describe the tumor phenotype based on mathematical algorithms. This new way of analyzing imaging data has become known as radiomics. There is growing evidence that radiomics may have the potential to serve as a cost-effective and non- invasive biomarker for personalized medicine by providing prognostic information regarding tumor stage, metastatic potential, treatment response and overall survival. Patients with early-stage NSCLC are particularly suitable to evaluate the prognostic impact of radiomics for non-invasive tumor profiling, as CT images and histopathological analysis are routinely available for these patients. Moreover, as mentioned above, despite successful surgical resection of the same disease stage it is well known but not predictable so far, that a certain subpopulation of these patients will experience recurrent disease indicating the need for precision medicine. Therefore, the purpose of the proposed study is to investigate the prognostic performance of radiomic data in patients with early-stage NSCLC with the aim to identify predictive image features to estimate the metastatic potential and overall survival. This is of great importance not only for personalized medicine but also from an interdisciplinary and socioeconomic point of view.